基于串行分类器的字符识别  

Character Recognition Based on the Cascading Classifier

作　　者：黄志斌[1] 陈锻生[1] 

机构地区：[1]华侨大学计算机科学系,福建泉州362011

出　　处：《计算机应用研究》2002年第12期43-46,共4页Application Research of Computers

摘　　要：在特征提取的基础上,比较了采用Bayes分类器和基于特征值的BP网络两种常用字符识别方案;最后在Bayes分类器的基础上,采用二级串行结构分类器融合的改进处理方案对车牌中的大写英文字母进行字符识别。实验结果表明,该方法达到较高的识别率,而且可很容易地推广到车牌识别中的汉字字符识别和数字字符识别相应的处理方案。Based on the step of feature-extracting, this paper compares two commonly used methods of Bayes classifer and BP network classifier, finally recognize English characters by combining the cascading classifier based on the method of Bayes classifier. Experimental results show that high correct accuracy can be obtained and this method can be easily adjusted to be the method of recognizing the Chinese characters and the numerals on the car plate.

关 键 词：串行分类器 字符识别 模式识别 特征提取 BP网络 人工神经网络 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]